1 /* $NetBSD: compress.c,v 1.6 2013/01/03 23:05:38 christos Exp $ */ 2 3 /* 4 * Copyright (c) Ian F. Darwin 1986-1995. 5 * Software written by Ian F. Darwin and others; 6 * maintained 1995-present by Christos Zoulas and others. 7 * 8 * Redistribution and use in source and binary forms, with or without 9 * modification, are permitted provided that the following conditions 10 * are met: 11 * 1. Redistributions of source code must retain the above copyright 12 * notice immediately at the beginning of the file, without modification, 13 * this list of conditions, and the following disclaimer. 14 * 2. Redistributions in binary form must reproduce the above copyright 15 * notice, this list of conditions and the following disclaimer in the 16 * documentation and/or other materials provided with the distribution. 17 * 18 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND 19 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 20 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 21 * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE FOR 22 * ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 23 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 24 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 25 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 26 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 27 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 28 * SUCH DAMAGE. 29 */ 30 /* 31 * compress routines: 32 * zmagic() - returns 0 if not recognized, uncompresses and prints 33 * information if recognized 34 * uncompress(method, old, n, newch) - uncompress old into new, 35 * using method, return sizeof new 36 */ 37 #include "file.h" 38 39 #ifndef lint 40 #if 0 41 FILE_RCSID("@(#)$File: compress.c,v 1.70 2012/11/07 17:54:48 christos Exp $") 42 #else 43 __RCSID("$NetBSD: compress.c,v 1.6 2013/01/03 23:05:38 christos Exp $"); 44 #endif 45 #endif 46 47 #include "magic.h" 48 #include <stdlib.h> 49 #ifdef HAVE_UNISTD_H 50 #include <unistd.h> 51 #endif 52 #include <string.h> 53 #include <errno.h> 54 #ifndef __MINGW32__ 55 #include <sys/ioctl.h> 56 #endif 57 #ifdef HAVE_SYS_WAIT_H 58 #include <sys/wait.h> 59 #endif 60 #if defined(HAVE_SYS_TIME_H) 61 #include <sys/time.h> 62 #endif 63 #if defined(HAVE_ZLIB_H) && defined(HAVE_LIBZ) 64 #define BUILTIN_DECOMPRESS 65 #include <zlib.h> 66 #endif 67 68 private const struct { 69 const char magic[8]; 70 size_t maglen; 71 const char *argv[3]; 72 int silent; 73 } compr[] = { 74 { "\037\235", 2, { "gzip", "-cdq", NULL }, 1 }, /* compressed */ 75 /* Uncompress can get stuck; so use gzip first if we have it 76 * Idea from Damien Clark, thanks! */ 77 { "\037\235", 2, { "uncompress", "-c", NULL }, 1 }, /* compressed */ 78 { "\037\213", 2, { "gzip", "-cdq", NULL }, 1 }, /* gzipped */ 79 { "\037\236", 2, { "gzip", "-cdq", NULL }, 1 }, /* frozen */ 80 { "\037\240", 2, { "gzip", "-cdq", NULL }, 1 }, /* SCO LZH */ 81 /* the standard pack utilities do not accept standard input */ 82 { "\037\036", 2, { "gzip", "-cdq", NULL }, 0 }, /* packed */ 83 { "PK\3\4", 4, { "gzip", "-cdq", NULL }, 1 }, /* pkzipped, */ 84 /* ...only first file examined */ 85 { "BZh", 3, { "bzip2", "-cd", NULL }, 1 }, /* bzip2-ed */ 86 { "LZIP", 4, { "lzip", "-cdq", NULL }, 1 }, 87 { "\3757zXZ\0",6,{ "xz", "-cd", NULL }, 1 }, /* XZ Utils */ 88 { "LRZI", 4, { "lrzip", "-dqo-", NULL }, 1 }, /* LRZIP */ 89 }; 90 91 #define NODATA ((size_t)~0) 92 93 private ssize_t swrite(int, const void *, size_t); 94 #if HAVE_FORK 95 private size_t ncompr = sizeof(compr) / sizeof(compr[0]); 96 private size_t uncompressbuf(struct magic_set *, int, size_t, 97 const unsigned char *, unsigned char **, size_t); 98 #ifdef BUILTIN_DECOMPRESS 99 private size_t uncompressgzipped(struct magic_set *, const unsigned char *, 100 unsigned char **, size_t); 101 #endif 102 103 protected int 104 file_zmagic(struct magic_set *ms, int fd, const char *name, 105 const unsigned char *buf, size_t nbytes) 106 { 107 unsigned char *newbuf = NULL; 108 size_t i, nsz; 109 int rv = 0; 110 int mime = ms->flags & MAGIC_MIME; 111 112 if ((ms->flags & MAGIC_COMPRESS) == 0) 113 return 0; 114 115 for (i = 0; i < ncompr; i++) { 116 if (nbytes < compr[i].maglen) 117 continue; 118 if (memcmp(buf, compr[i].magic, compr[i].maglen) == 0 && 119 (nsz = uncompressbuf(ms, fd, i, buf, &newbuf, 120 nbytes)) != NODATA) { 121 ms->flags &= ~MAGIC_COMPRESS; 122 rv = -1; 123 if (file_buffer(ms, -1, name, newbuf, nsz) == -1) 124 goto error; 125 126 if (mime == MAGIC_MIME || mime == 0) { 127 if (file_printf(ms, mime ? 128 " compressed-encoding=" : " (") == -1) 129 goto error; 130 } 131 132 if ((mime == 0 || mime & MAGIC_MIME_ENCODING) && 133 file_buffer(ms, -1, NULL, buf, nbytes) == -1) 134 goto error; 135 136 if (!mime && file_printf(ms, ")") == -1) 137 goto error; 138 rv = 1; 139 break; 140 } 141 } 142 error: 143 free(newbuf); 144 ms->flags |= MAGIC_COMPRESS; 145 return rv; 146 } 147 #endif 148 /* 149 * `safe' write for sockets and pipes. 150 */ 151 private ssize_t 152 swrite(int fd, const void *buf, size_t n) 153 { 154 ssize_t rv; 155 size_t rn = n; 156 157 do 158 switch (rv = write(fd, buf, n)) { 159 case -1: 160 if (errno == EINTR) 161 continue; 162 return -1; 163 default: 164 n -= rv; 165 buf = CAST(const char *, buf) + rv; 166 break; 167 } 168 while (n > 0); 169 return rn; 170 } 171 172 173 /* 174 * `safe' read for sockets and pipes. 175 */ 176 protected ssize_t 177 sread(int fd, void *buf, size_t n, int canbepipe __attribute__((__unused__))) 178 { 179 ssize_t rv; 180 #ifdef FIONREAD 181 int t = 0; 182 #endif 183 size_t rn = n; 184 185 if (fd == STDIN_FILENO) 186 goto nocheck; 187 188 #ifdef FIONREAD 189 if (canbepipe && (ioctl(fd, FIONREAD, &t) == -1 || t == 0)) { 190 #ifdef FD_ZERO 191 ssize_t cnt; 192 for (cnt = 0;; cnt++) { 193 fd_set check; 194 struct timeval tout = {0, 100 * 1000}; 195 int selrv; 196 197 FD_ZERO(&check); 198 FD_SET(fd, &check); 199 200 /* 201 * Avoid soft deadlock: do not read if there 202 * is nothing to read from sockets and pipes. 203 */ 204 selrv = select(fd + 1, &check, NULL, NULL, &tout); 205 if (selrv == -1) { 206 if (errno == EINTR || errno == EAGAIN) 207 continue; 208 } else if (selrv == 0 && cnt >= 5) { 209 return 0; 210 } else 211 break; 212 } 213 #endif 214 (void)ioctl(fd, FIONREAD, &t); 215 } 216 217 if (t > 0 && (size_t)t < n) { 218 n = t; 219 rn = n; 220 } 221 #endif 222 223 nocheck: 224 do 225 switch ((rv = read(fd, buf, n))) { 226 case -1: 227 if (errno == EINTR) 228 continue; 229 return -1; 230 case 0: 231 return rn - n; 232 default: 233 n -= rv; 234 buf = ((char *)buf) + rv; 235 break; 236 } 237 while (n > 0); 238 return rn; 239 } 240 241 protected int 242 file_pipe2file(struct magic_set *ms, int fd, const void *startbuf, 243 size_t nbytes) 244 { 245 char buf[4096]; 246 ssize_t r; 247 int tfd; 248 249 (void)strlcpy(buf, "/tmp/file.XXXXXX", sizeof buf); 250 #ifndef HAVE_MKSTEMP 251 { 252 char *ptr = mktemp(buf); 253 tfd = open(ptr, O_RDWR|O_TRUNC|O_EXCL|O_CREAT, 0600); 254 r = errno; 255 (void)unlink(ptr); 256 errno = r; 257 } 258 #else 259 { 260 int te; 261 tfd = mkstemp(buf); 262 te = errno; 263 (void)unlink(buf); 264 errno = te; 265 } 266 #endif 267 if (tfd == -1) { 268 file_error(ms, errno, 269 "cannot create temporary file for pipe copy"); 270 return -1; 271 } 272 273 if (swrite(tfd, startbuf, nbytes) != (ssize_t)nbytes) 274 r = 1; 275 else { 276 while ((r = sread(fd, buf, sizeof(buf), 1)) > 0) 277 if (swrite(tfd, buf, (size_t)r) != r) 278 break; 279 } 280 281 switch (r) { 282 case -1: 283 file_error(ms, errno, "error copying from pipe to temp file"); 284 return -1; 285 case 0: 286 break; 287 default: 288 file_error(ms, errno, "error while writing to temp file"); 289 return -1; 290 } 291 292 /* 293 * We duplicate the file descriptor, because fclose on a 294 * tmpfile will delete the file, but any open descriptors 295 * can still access the phantom inode. 296 */ 297 if ((fd = dup2(tfd, fd)) == -1) { 298 file_error(ms, errno, "could not dup descriptor for temp file"); 299 return -1; 300 } 301 (void)close(tfd); 302 if (lseek(fd, (off_t)0, SEEK_SET) == (off_t)-1) { 303 file_badseek(ms); 304 return -1; 305 } 306 return fd; 307 } 308 #if HAVE_FORK 309 #ifdef BUILTIN_DECOMPRESS 310 311 #define FHCRC (1 << 1) 312 #define FEXTRA (1 << 2) 313 #define FNAME (1 << 3) 314 #define FCOMMENT (1 << 4) 315 316 private size_t 317 uncompressgzipped(struct magic_set *ms, const unsigned char *old, 318 unsigned char **newch, size_t n) 319 { 320 unsigned char flg = old[3]; 321 size_t data_start = 10; 322 z_stream z; 323 int rc; 324 325 if (flg & FEXTRA) { 326 if (data_start+1 >= n) 327 return 0; 328 data_start += 2 + old[data_start] + old[data_start + 1] * 256; 329 } 330 if (flg & FNAME) { 331 while(data_start < n && old[data_start]) 332 data_start++; 333 data_start++; 334 } 335 if(flg & FCOMMENT) { 336 while(data_start < n && old[data_start]) 337 data_start++; 338 data_start++; 339 } 340 if(flg & FHCRC) 341 data_start += 2; 342 343 if (data_start >= n) 344 return 0; 345 if ((*newch = CAST(unsigned char *, malloc(HOWMANY + 1))) == NULL) { 346 return 0; 347 } 348 349 /* XXX: const castaway, via strchr */ 350 z.next_in = (Bytef *)strchr((const char *)old + data_start, 351 old[data_start]); 352 z.avail_in = CAST(uint32_t, (n - data_start)); 353 z.next_out = *newch; 354 z.avail_out = HOWMANY; 355 z.zalloc = Z_NULL; 356 z.zfree = Z_NULL; 357 z.opaque = Z_NULL; 358 359 /* LINTED bug in header macro */ 360 rc = inflateInit2(&z, -15); 361 if (rc != Z_OK) { 362 file_error(ms, 0, "zlib: %s", z.msg); 363 return 0; 364 } 365 366 rc = inflate(&z, Z_SYNC_FLUSH); 367 if (rc != Z_OK && rc != Z_STREAM_END) { 368 file_error(ms, 0, "zlib: %s", z.msg); 369 return 0; 370 } 371 372 n = (size_t)z.total_out; 373 (void)inflateEnd(&z); 374 375 /* let's keep the nul-terminate tradition */ 376 (*newch)[n] = '\0'; 377 378 return n; 379 } 380 #endif 381 382 private size_t 383 uncompressbuf(struct magic_set *ms, int fd, size_t method, 384 const unsigned char *old, unsigned char **newch, size_t n) 385 { 386 int fdin[2], fdout[2]; 387 ssize_t r; 388 pid_t pid; 389 390 #ifdef BUILTIN_DECOMPRESS 391 /* FIXME: This doesn't cope with bzip2 */ 392 if (method == 2) 393 return uncompressgzipped(ms, old, newch, n); 394 #endif 395 (void)fflush(stdout); 396 (void)fflush(stderr); 397 398 if ((fd != -1 && pipe(fdin) == -1) || pipe(fdout) == -1) { 399 file_error(ms, errno, "cannot create pipe"); 400 return NODATA; 401 } 402 switch (pid = fork()) { 403 case 0: /* child */ 404 (void) close(0); 405 if (fd != -1) { 406 if (dup(fd) == -1) 407 _exit(1); 408 (void) lseek(0, (off_t)0, SEEK_SET); 409 } else { 410 if (dup(fdin[0]) == -1) 411 _exit(1); 412 (void) close(fdin[0]); 413 (void) close(fdin[1]); 414 } 415 416 (void) close(1); 417 if (dup(fdout[1]) == -1) 418 _exit(1); 419 (void) close(fdout[0]); 420 (void) close(fdout[1]); 421 #ifndef DEBUG 422 if (compr[method].silent) 423 (void)close(2); 424 #endif 425 426 (void)execvp(compr[method].argv[0], 427 (char *const *)(intptr_t)compr[method].argv); 428 #ifdef DEBUG 429 (void)fprintf(stderr, "exec `%s' failed (%s)\n", 430 compr[method].argv[0], strerror(errno)); 431 #endif 432 exit(1); 433 /*NOTREACHED*/ 434 case -1: 435 file_error(ms, errno, "could not fork"); 436 return NODATA; 437 438 default: /* parent */ 439 (void) close(fdout[1]); 440 if (fd == -1) { 441 (void) close(fdin[0]); 442 /* 443 * fork again, to avoid blocking because both 444 * pipes filled 445 */ 446 switch (fork()) { 447 case 0: /* child */ 448 (void)close(fdout[0]); 449 if (swrite(fdin[1], old, n) != (ssize_t)n) { 450 #ifdef DEBUG 451 (void)fprintf(stderr, 452 "Write failed (%s)\n", 453 strerror(errno)); 454 #endif 455 exit(1); 456 } 457 exit(0); 458 /*NOTREACHED*/ 459 460 case -1: 461 #ifdef DEBUG 462 (void)fprintf(stderr, "Fork failed (%s)\n", 463 strerror(errno)); 464 #endif 465 exit(1); 466 /*NOTREACHED*/ 467 468 default: /* parent */ 469 break; 470 } 471 (void) close(fdin[1]); 472 fdin[1] = -1; 473 } 474 475 if ((*newch = (unsigned char *) malloc(HOWMANY + 1)) == NULL) { 476 #ifdef DEBUG 477 (void)fprintf(stderr, "Malloc failed (%s)\n", 478 strerror(errno)); 479 #endif 480 n = 0; 481 goto err; 482 } 483 if ((r = sread(fdout[0], *newch, HOWMANY, 0)) <= 0) { 484 #ifdef DEBUG 485 (void)fprintf(stderr, "Read failed (%s)\n", 486 strerror(errno)); 487 #endif 488 free(*newch); 489 n = 0; 490 newch[0] = '\0'; 491 goto err; 492 } else { 493 n = r; 494 } 495 /* NUL terminate, as every buffer is handled here. */ 496 (*newch)[n] = '\0'; 497 err: 498 if (fdin[1] != -1) 499 (void) close(fdin[1]); 500 (void) close(fdout[0]); 501 #ifdef WNOHANG 502 while (waitpid(pid, NULL, WNOHANG) != -1) 503 continue; 504 #else 505 (void)wait(NULL); 506 #endif 507 (void) close(fdin[0]); 508 509 return n; 510 } 511 } 512 #endif 513